July 8, 2025 | TechXNow Newsroom
As artificial intelligence tools evolve at lightning speed, many companies find themselves struggling to adapt. One such company—Brex, the corporate credit card startup—is tackling the problem by fundamentally rethinking how it approaches software procurement.
The Challenge: Slow Procurement in a Fast-Moving World
At the HumanX AI Conference this past March, Brex CTO James Reggio explained how traditional procurement models were holding the company back. The startup initially attempted to adopt AI tools through its usual months-long piloting process.
“The procurement process itself would actually run so long that the teams… lost interest in the tool by the time we got through all the necessary internal controls,” said Reggio.
This mismatch between AI’s rapid development and Brex’s internal timelines made it clear: a new approach was necessary.
The Pivot: Building a Faster, Smarter Framework
Brex responded by reworking its procurement process from the ground up. The company introduced a streamlined framework for data processing agreements and legal validations, enabling faster onboarding of AI tools.
This new method helped Brex quickly get tools into the hands of testers, accelerating the feedback loop and allowing the most useful tools to surface faster.
Testing for Real Value: The “Superhuman Product-Market-Fit” Approach
To decide which tools to scale beyond pilot testing, Brex uses what Reggio calls a “superhuman product-market-fit test.” This process focuses on identifying tools that are uniquely valuable to employees, based on actual usage and impact.
“We go deep with the folks who are getting the most value… to figure out whether it is actually unique enough to retain,” Reggio noted.
Since adopting this strategy, Brex has tried over 1,000 AI tools, but has also canceled or declined to renew several large deployments—showing its commitment to utility over hype.
Empowering Engineers with Micro-Budgets
A key piece of Brex’s new model is employee-level autonomy. Engineers are given a $50 monthly budget to license AI software tools of their choice from a pre-approved list.
“By delegating that spending authority… they make optimal decisions for optimizing their workflows,” said Reggio.
Interestingly, this decentralized approach hasn’t led to a single dominant tool. Instead, different teams choose tools that best suit their specific needs—proving that flexibility beats uniformity in fast-moving environments.
Smarter Licensing Decisions from Real Usage Data
This engineer-led experimentation also feeds into smarter decisions at the organizational level. By tracking which tools are used most, Brex can identify when it makes sense to pursue broader licensing agreements, based on actual user adoption rather than assumptions.
Key Takeaway: Embrace the Messiness
Reggio’s advice to other enterprises? Don’t try to perfect the process. Instead, embrace the chaos of the current AI cycle.
“Knowing that you’re not going to always make the right decision out of the gate is paramount,” he said. “The one mistake we could make is overthinking this for six to nine months—and getting left behind.”
Brex’s experience offers a valuable blueprint for how modern companies can stay ahead in a chaotic, rapidly evolving AI landscape. By prioritizing speed, flexibility, and user-led experimentation, the company is proving that adapting quickly may matter more than getting it perfect.
Stay tuned with TechXNow for more insights on how forward-thinking companies are navigating the AI revolution
sources ( Techcrunch )
